{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:7EY5K3PMPHQKYRQ3SJ3PJITHDW","short_pith_number":"pith:7EY5K3PM","schema_version":"1.0","canonical_sha256":"f931d56dec79e0ac461b9276f4a2671d98fe257f1fe6c4c6704fe7cd437dab7e","source":{"kind":"arxiv","id":"2106.00545","version":3},"attestation_state":"computed","paper":{"title":"Counterfactual Invariance to Spurious Correlations: Why and How to Pass Stress Tests","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexander D'Amour, Jacob Eisenstein, Steve Yadlowsky, Victor Veitch","submitted_at":"2021-05-31T14:39:38Z","abstract_excerpt":"Informally, a 'spurious correlation' is the dependence of a model on some aspect of the input data that an analyst thinks shouldn't matter. In machine learning, these have a know-it-when-you-see-it character; e.g., changing the gender of a sentence's subject changes a sentiment predictor's output. To check for spurious correlations, we can 'stress test' models by perturbing irrelevant parts of input data and seeing if model predictions change. In this paper, we study stress testing using the tools of causal inference. We introduce counterfactual invariance as a formalization of the requirement"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2106.00545","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-05-31T14:39:38Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"201b05197e9883b81023c6d8e8dfb6344c32686c36ea95f8bcc9052f765ef5e8","abstract_canon_sha256":"a7f71a3d11f714f3677897aed5a19aad204151ae5aca0a0da7aebc94fc6c68a4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:28:37.487340Z","signature_b64":"h/n2uBcfj5ughbqOiAR5C9umU4+ZlCkLg/PnsAiP4VfuLSBx5sOVHkgAwr7FdjYUy2GSSkxOwpugqcu/gIjDBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f931d56dec79e0ac461b9276f4a2671d98fe257f1fe6c4c6704fe7cd437dab7e","last_reissued_at":"2026-07-05T03:28:37.486893Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:28:37.486893Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Counterfactual Invariance to Spurious Correlations: Why and How to Pass Stress Tests","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexander D'Amour, Jacob Eisenstein, Steve Yadlowsky, Victor Veitch","submitted_at":"2021-05-31T14:39:38Z","abstract_excerpt":"Informally, a 'spurious correlation' is the dependence of a model on some aspect of the input data that an analyst thinks shouldn't matter. In machine learning, these have a know-it-when-you-see-it character; e.g., changing the gender of a sentence's subject changes a sentiment predictor's output. To check for spurious correlations, we can 'stress test' models by perturbing irrelevant parts of input data and seeing if model predictions change. In this paper, we study stress testing using the tools of causal inference. We introduce counterfactual invariance as a formalization of the requirement"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.00545","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2106.00545/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2106.00545","created_at":"2026-07-05T03:28:37.486964+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.00545v3","created_at":"2026-07-05T03:28:37.486964+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.00545","created_at":"2026-07-05T03:28:37.486964+00:00"},{"alias_kind":"pith_short_12","alias_value":"7EY5K3PMPHQK","created_at":"2026-07-05T03:28:37.486964+00:00"},{"alias_kind":"pith_short_16","alias_value":"7EY5K3PMPHQKYRQ3","created_at":"2026-07-05T03:28:37.486964+00:00"},{"alias_kind":"pith_short_8","alias_value":"7EY5K3PM","created_at":"2026-07-05T03:28:37.486964+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08317","citing_title":"Blind-Spots-Bench: Evaluating Blind Spots in Multimodal Models","ref_index":74,"is_internal_anchor":true},{"citing_arxiv_id":"2601.21350","citing_title":"Factored Causal Representation Learning for Robust Reward Modeling in RLHF","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17957","citing_title":"Contextualized Code Pretraining for Code Generation","ref_index":44,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW","json":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW.json","graph_json":"https://pith.science/api/pith-number/7EY5K3PMPHQKYRQ3SJ3PJITHDW/graph.json","events_json":"https://pith.science/api/pith-number/7EY5K3PMPHQKYRQ3SJ3PJITHDW/events.json","paper":"https://pith.science/paper/7EY5K3PM"},"agent_actions":{"view_html":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW","download_json":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW.json","view_paper":"https://pith.science/paper/7EY5K3PM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.00545&json=true","fetch_graph":"https://pith.science/api/pith-number/7EY5K3PMPHQKYRQ3SJ3PJITHDW/graph.json","fetch_events":"https://pith.science/api/pith-number/7EY5K3PMPHQKYRQ3SJ3PJITHDW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW/action/storage_attestation","attest_author":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW/action/author_attestation","sign_citation":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW/action/citation_signature","submit_replication":"https://pith.science/pith/7EY5K3PMPHQKYRQ3SJ3PJITHDW/action/replication_record"}},"created_at":"2026-07-05T03:28:37.486964+00:00","updated_at":"2026-07-05T03:28:37.486964+00:00"}